{ "s_eT_tgw1.0_wr0.2": { "test_acc": 0.5605, "mean_bp_cosine": 0.46006373316049576, "mean_rho": 0.5312944203615189, "mean_nudge": -0.03811332583427429, "bp_cosine_per_layer": [ 0.18216726183891296, 0.5338927507400513, 0.5597370862960815, 0.5644578337669373 ], "rho_per_layer": [ 0.2443590760231018, 0.5769795775413513, 0.6636242270469666, 0.6402148008346558 ], "final_value_loss": 0.39113953742980956, "s_type": "eT", "term_grad_weight": 1.0, "warmup_ratio": 0.2 }, "s_deltaL_tgw1.0_wr0.2": { "test_acc": 0.442, "mean_bp_cosine": 0.5583586767315865, "mean_rho": 0.49782785028219223, "mean_nudge": -0.051266513764858246, "bp_cosine_per_layer": [ 0.2746746242046356, 0.6255925893783569, 0.6661019921302795, 0.6670655012130737 ], "rho_per_layer": [ 0.27920976281166077, 0.5733566284179688, 0.594862163066864, 0.5438828468322754 ], "final_value_loss": 0.47215704221725463, "s_type": "deltaL", "term_grad_weight": 1.0, "warmup_ratio": 0.2 }, "s_eT_hL_tgw1.0_wr0.2": { "test_acc": 0.559, "mean_bp_cosine": 0.0016190913447644562, "mean_rho": 0.016410242416895926, "mean_nudge": -0.00021975510026095435, "bp_cosine_per_layer": [ -0.004644347354769707, 0.0010294992243871093, -0.0010337075218558311, 0.011124921031296253 ], "rho_per_layer": [ 0.046507176011800766, 0.007014757487922907, -0.006595376878976822, 0.018714413046836853 ], "final_value_loss": 0.8668150995254517, "s_type": "eT_hL", "term_grad_weight": 1.0, "warmup_ratio": 0.2 }, "s_deltaL_hL_tgw1.0_wr0.2": { "test_acc": 0.5635, "mean_bp_cosine": 0.018434561206959188, "mean_rho": 0.026365829166024923, "mean_nudge": -0.0017427445782232098, "bp_cosine_per_layer": [ 0.013983936980366707, 0.02003202959895134, 0.00282791955396533, 0.036894358694553375 ], "rho_per_layer": [ -0.006572568789124489, 0.049806319177150726, 0.020004384219646454, 0.042225182056427 ], "final_value_loss": 0.7973948855400086, "s_type": "deltaL_hL", "term_grad_weight": 1.0, "warmup_ratio": 0.2 } }